[Docs] Align documentation with tirx/s_tir namespace split

[tlopex]

Since we've split the old tir namespace into tirx (core IR / lowering) and s_tir (schedule primitives / auto-tuning), some outdated documentation need to be updated. The global rename still leaves a few concept-level references using "tirx" in prose (for example, "Relax and tirx programs"). Since "tirx" now refers only to one part of the old TensorIR stack, these higher-level references should use "TensorIR" instead, so they correctly cover both tirx and s_tir.

In this PR, we

• Add tirx / s_tir module descriptions to docs/deep_dive/tensor_ir/index.rst and docs/arch/index.rst (new tvm/s_tir section, updated tvm/tirx section).
• Fix concept-level prose in docs/arch/index.rst and docs/arch/pass_infra.rst to use "TensorIR" instead of "tirx" where referring to the concept rather than the namespace.
• Fix docs/arch/runtimes/vulkan.rst to use "TIR" instead of "tirx" in debug shader description.
• Correct tvm/dlight → tvm/s_tir/dlight section path and "tirx schedules" → "s_tir schedules" in docs/arch/index.rst.
• Revert unintended label changes in abstraction.rst and tir_creation.py (labels kept as _tir-abstraction, _tir-creation).
• Revert unintended title change in tir_transformation.py (kept as "Transformation").
• Revert exclude-members change in tirx/tirx.rst (kept original list).

[tlopex]

cc @tqchen

[gemini-code-assist]

Code Review

This pull request updates the documentation to reflect the architectural split of the tir module into tirx (core IR and lowering) and s_tir (scheduling and tuning). It replaces internal terminology with more descriptive terms like 'TensorIR' and 'TIR' across several architecture and runtime guides. A suggestion was made to clarify the TVMScript entry point description to better explain how users interact with these modules.

[gemini-code-assist]

For better clarity, this sentence could be rephrased to emphasize that tirx is the user-facing entry point in TVMScript. The current phrasing 'both modules are accessed via' might be slightly confusing as it's not immediately clear how s_tir components are accessed through the tirx import.

Suggested change

	In TVMScript, both modules are accessed via
	``from tvm.script import tirx as T``.
	In TVMScript, the `tirx` Python module is the main entrypoint for
	writing TensorIR functions, and is typically imported as `T`: ``from tvm.script import tirx as T``.